Residual Concatenated Network for ODBTC Image Restoration

Abstract

This paper proposes Residual Concatenated Network (RCN) for improving the quality of Ordered Dither Block Truncation Coding (ODBTC) decoded image. This method inherits the effectiveness of Convolutional Neural Networks (CNN) for suppressing the impulsive noise occurred in decoded image. It suppresses the noise by applying a series of convolutional operations. The network directly performs learning process via an end-to-end mapping approach. The experimental results reveal that the proposed approach yields a promising result in the ODBTC image restoration.